Scottish voting intentions for general election (Ipsos-Mori):
Liberal Democrats 11%
Seat projection: SNP 48 (+13), Conservatives 6 (-7), Liberal Democrats 4 (n/c), Labour 1 (-6)
Should Scotland be an independent country?
The seat projection is obviously markedly better for the SNP, and worse for the Tories, than we saw from the YouGov model last night. So which is likely to be closer to the truth? The YouGov projection is more sophisticated, because it takes account of different trends in different areas - for example that there may be more of a swing against the Tories in Remain-friendly East Renfrewshire than in Leave-friendly Banff & Buchan. By contrast, the projection from the Ipsos-Mori poll is very crude and based on assumptions of a uniform swing. But the fundamental reason why the SNP would get more seats if Ipsos-Mori are right is that the poll shows the party with a higher share of the vote than YouGov detected. If it was possible to run a subtler projection model based on Ipsos-Mori's figures, you'd probably still see the SNP gaining more Tory seats than YouGov predicted.
So it really just boils down to whether you think Ipsos-Mori or YouGov are estimating the vote shares of each party more accurately. There was a time when we'd have assumed that Ipsos-Mori's data collection method by telephone was bound to produce more accurate results than YouGov's online approach, but what happened in the EU referendum raised a few question marks over that.
Incidentally, the fieldwork dates don't explain the difference - the Ipsos-Mori poll was conducted over roughly the same period as the fieldwork for the YouGov model.
More to follow...
I have four more constituency previews in today's edition of The National - this time it's East Lothian, Coatbridge, Chryston & Bellshill, Glasgow South and Glasgow Central.